I was wondering about the possibility of using the following setup with write-caching

btrfs
    bcache
        fast
            mdadm (RAID-1)
                ssd_1
                ssd_2
        slow
            mdadm (RAID-5)
                hdd_1
                hdd_2
                hdd_3
                hdd_4

Is this viable / reasonable? The arch wiki mentions the possibility of data loss when using write-caching when the SSD fails, but shouldn’t the SSD RAID array prevent that? It also mentions “bcache and BTRFS could leave you with a corrupted file system” is this still true? The wiki page mentions that it’s unclear if this is still an issue.

Someone also left the following comment on the discussions page regarding BTRFS in 2023:

The issues with btrfs + bcache were fixed 10 years ago. The btrfs wiki no longer mentions historic gotchas for kernels older than 4.14. I think we should remove this warning. Any objections?

  • @[email protected]English
    1•22 hours ago

    I’ve got a similar setup without the raid bcache cache drive and wish I did that as it should make maintenance easier. I think it should work. Tho I dunno the gotchas for btrfs and have only done it with xfs. Why btrfs if you don’t mind me asking?

    • qazOPEnglish
      1•17 hours ago

      I would like to use BTRFS for deduplication, CoW, and snapshots.

  • @[email protected]English
    2•1 day ago

    Does this use Btfs’ RAID5? If so you might want to avoid since RAID5/6 arent production ready for Btrfs and contain known bugs that can lead to parity loss.

    • qazOPEnglish
      2•1 day ago

      No, I’m aware of BTRFS’s RAID 5/6 issues, this would use mdadm’s RAID with BTRFS on the bcache block device.